Explainable Machine Learning Based Optimization of Strength, Durability, and Carbon Efficiency of Fly Ash–GGBS Geopolymer Concrete

Geopolymer concrete (GPC) is widely recognized as a sustainable alternative to ordinary Portland cement concrete; however, optimizing long-term durability, mechanical performance, and environmental impact remains a critical challenge. This study presents an integrated experimental, machine learning (ML), and life cycle assessment (LCA) framework to optimize the strength, durability, and carbon trade-off of fly ash (FA) and ground granulated blast furnace slag (GGBS) based GPC. Sixteen geopolymer mixtures were designed by systematically varying sodium hydroxide molarity (8–14 M), Si/Al ratio, FA:GGBS ratio (70:30 and 50:50), and curing regime (ambient curing and heat curing at 60 ⁰ C). Compressive strength, rapid chloride penetration (RCP), water absorption, and sorptivity were evaluated at 28, 90, and 180 days to assess long-term performance. The results revealed continuous strength development and progressive durability enhancement with curing age, with an optimum NaOH molarity of 12 M identified for both mechanical and transport properties. GGBS rich mixtures exhibited superior long-term performance due to enhanced matrix densification, while heat curing significantly improved early-age strength and durability. The highest 180-day compressive strength of 75.1 MPa was obtained for the GPC-12-50-H mixture, whereas ambient-cured GPC-12-50-A demonstrated comparable long-term performance with improved sustainability. ML models, including Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Artificial Neural Networks (ANN), were developed to predict compressive strength. XGBoost achieved the highest prediction accuracy (R² = 0.98), followed by RF (R² = 0.97). Explainable ML using SHAP analysis identified curing regime, NaOH molarity, and Si/Al ratio as the most influential parameters. Life cycle assessment showed increasing global warming potential and embodied energy with higher alkalinity, GGBS content, and heat curing. A carbon efficiency index based on 180 day strength identified a 12 M NaOH and 50:50 (FA:GGBS) ratio as the most sustainable mix design.